#emc-devel | Logs for 2010-05-21

[00:02:37] <andypugh> http://www.youtube.com/watch?v=96dWOEa4Djs
[00:08:50] <SWPadnos> I think I've seen that video before
[00:09:21] <andypugh> Yeah, it's fairly old. But shows the tech works.
[00:09:35] <SWPadnos> they have several RAID controllers in their parts list, I wonder which one(s) they ended up using
[00:10:30] <SWPadnos> ah, ok. they used an ARECA 24-port RAID card (which didn't do as well as they thought it would), so they added an 8-port Adaptec card
[00:11:22] <SWPadnos> and then ended up with 10 on the Areca card, 8 on the Adaptec, and 6 direct to the motherboard
[00:11:43] <SWPadnos> so 3 PCIe interfaces were used to get the 2.1GB/s read speed
[00:46:40] <JT-Hardinge> is it normal for F in the MDI window to be -1 after running a program?
[00:47:52] <JT-Hardinge> if you try and make a feed move you get cannot do g1 with zero feed rate and even after you set a feed rate it will not move until you leave the MDI window and come back to it
[00:47:55] <JT-Hardinge> 2.5
[00:48:26] <JT-Hardinge> bbl
[02:16:17] <CIA-2> EMC: 03cmorley 07v2.4_branch * r4d05fc3ecf25 10/src/emc/usr_intf/pncconf/pncconf-help/axisdiagram1.png: Remove text from diagram - its hard to read
[11:24:49] <morficmobile> the rtai kernel for 10.4 someone is working on, what kernel version is it based on?
[11:28:57] <morficmobile> mozmck_work: if you get this much jitter, i wonder how bfs would play with rtai, if it applies at all it might be interesting, considering i got RT benefits with bfs patched kernel in jackd
[11:38:14] <alex_joni> 2.6.32.x
[11:39:01] <morficmobile> thanks
[13:28:14] <micges_work> jepler: it seems that stepgen enabling issue is fixed
[13:29:36] <micges_work> in old code every 8 times there was glitch, after that patch there wasn't any glitch in 6 tests
[13:42:02] <mozmck> I should have mentioned that the jitter on that computer was just about the same with 8.04.
[16:37:23] <Dave911> Where does the logger run?
[16:39:01] <jepler> same place that does linuxcnc.org web hosting
[16:39:47] <Dave911> oh ... thanks for the tips ... I'll let you know what I find and supply a patch if this works out ..
[17:52:20] <mozmck_work> hmm, so far it looks like a kernel compiled for single core on this single core machine is running with less jitter than the SMP kernel
[17:52:46] <mozmck_work> ~20000 instead of ~36000
[17:53:01] <mozmck_work> oops, just jumped to 25000
[18:05:30] <jepler> mozmck_work: is the "kernel won't boot (reliably)" problem solved now?
[18:06:26] <mozmck_work> I haven't had a single problem since downgrading to grub-legacy
[18:06:45] <mozmck_work> Several folks on the rtai list are reporting the same thing
[18:06:48] <jepler> I wonder if the live cd installer will work if that is the bootloader
[18:07:07] <mozmck_work> I'm about to find out, hopefully this weekend.
[18:07:09] <jepler> I am starting to think about whether a 10.04 + 2.4.x live cd in time for cnc workshop is a reasonable goal or not
[18:07:22] <mozmck_work> when is the workshop?
[18:07:33] <jepler> about a month from now, I think
[18:07:57] <mozmck_work> oh, I hope to have a livecd ready for some testing by the end of next week anyhow.
[18:08:19] <mozmck_work> The main hitch might be grub
[18:09:34] <mozmck_work> There are also reports that using CONFIG_PREEMPT_INVOLUNTARY and not CONFIG_PREEMPT allows the kernel to boot with grub2, but I need to test that some more.
[18:09:36] <jepler> I figured you'd get the kernel more widely tested before tackling the live cd
[18:10:05] <mozmck_work> I'm going to try and get a live cd built, and it is fairly trivial to update the kernel in it.
[18:12:31] <jepler> at least for my 'net connection, what's not trivial is uploading it to linuxcnc for others to test
[18:12:48] <jepler> at least a kernel is only 20% as big as a whole live cd or something
[18:12:53] <mozmck_work> I don't know if the non-smp kernel will give better latency results on all single-core computers or not. I have 2 computers at home (maybe 3) that I plan to test it on.
[18:13:32] <mozmck_work> That's true. I wasn't planning on uploading a live cd until the kernel was more widely tested.
[18:14:31] <mozmck_work> But I can have the system ready to make into an image on my computer and even do some testing here so it will be a quick thing to put the final kernel in.
[18:16:11] <mozmck_work> At least on a 2.4 ghz dell computer we tried an SMP ubuntu 9.10 based kernel, and got latency around 10000. Didn't try a non-SMP kernel on it though.
[18:20:00] <jepler> if it runs, I'd be tempted to say go with SMP even if UP gives better latencies on some systems sometimes (but not if there are many systems it won't run at all on). SMP systems are a lot more prevalent now than they were 2 years ago.
[18:21:24] <mozmck_work> yeah, that's what I was wondering. I can also easily make a non-SMP kernel and packages available if someone needs the better latency
[18:21:49] <jepler> no, I don't want to be committed to building more emc packages. you get one (count it, one) kernel.
[18:22:16] <mozmck_work> Looks like the max jitter on this machine is about 28000 with glxgears running, and doing a clean build and packaging of emc2 at the same time.
[18:22:27] <mozmck_work> :) ok.
[18:22:54] <mozmck_work> so the difference on this machine is around 8us, but this one may not be typical.
[18:23:10] <jepler> how does that compare to the 8.04 kernel?
[18:23:46] <jepler> hi seb
[18:24:30] <seb_kuzminsky> hey there :-)
[18:25:43] <mozmck_work> jepler: the 8.04 kernel may have actually been closer to the 10.04 SMP
[18:26:01] <mozmck_work> I wrote it down but it's at home. I'll look later.
[18:26:17] <seb_kuzminsky> last night i plugged in a "logitech cordless rumblepad 2" and started turning it into a jog pendant :-)
[18:26:18] <seb_kuzminsky> http://www.amazon.com/Logitech-Cordless-Rumblepad-Vibration-Feedback/dp/B0002XJBGE
[18:26:28] <seb_kuzminsky> seems to be working fine
[18:28:06] <jepler> it's all in good fun until the batteries go dead in the middle of a jog
[18:28:32] <SWPadnos> save $5!! eliminate wireless!!! http://www.newegg.com/Product/Product.aspx?Item=N82E16826127207
[19:53:20] <JT-Work_> JT-Work_ is now known as JT-Work
[20:09:10] <SWPadnos> seb_kuzminsky, what do you see in HAL regarding the force feedback?
[20:09:21] <SWPadnos> are there output analog/digital pins that make the thing move?
[20:25:17] <jepler> SWPadnos: no, hal_input doesn't support force feedback
[20:25:30] <SWPadnos> heh, bumme
[20:25:31] <SWPadnos> r
[20:25:54] <SWPadnos> I wasn't sure if the force feedback would be listed as analog outputs or something
[20:28:32] <jepler> though actually a "cat /proc/bus/input/devices" from the device would be interesting to me. in theory the linux input layer supports force feedback but I don't understand it enough to try to add support in hal_input
[20:29:04] <jepler> or send me one or bring it to fest
[20:29:14] <SWPadnos> heh, yeah
[20:29:44] <SWPadnos> of course, at $25 it's not all that bad to just buy one to experiment with
[21:03:53] <alex_joni> http://www.dealextreme.com/details.dx/sku.618
[21:06:38] <SWPadnos> cheep
[21:07:47] <jepler> that's the second item I've read on dealextreme today where the review said the included drivers had a virus
[21:07:53] <jepler> the earlier one was http://www.dealextreme.com/details.dx/sku.16921
[21:08:31] <alex_joni> well.. I guess you won't mind that